AI is showing very different results depending on who is using it and how. Two studies by Jing Liu at the University of Maryland found that an AI math tutor called Goblins helped about 6,000 middle schoolers improve their test scores, particularly English learners and students with disabilities. Accelerate, a nonprofit research group, connected those gains to roughly five months of extra learning. However, the same research showed that a University of Maryland AI assistant hurt college students, lowering grades by about 0.11 points and reducing engagement. Only 15 percent of those college students actually used the tool. First-generation college students felt the negative effects most strongly.
On the security side, Maple Networks CTO Graham Tetley warned at International Cyber Expo 2026 that employees may accidentally share sensitive business data with AI platforms. He stressed that organizations need better visibility into how workers interact with AI tools. Maple Networks offers 24/7 Security Operations Centre services and CREST-accredited penetration testing to help businesses address these risks. Tetley said preventing unintended data exposure will stay a top priority for security teams.
Small businesses face similar AI security challenges. An interview with Simon Heath of The Final Step pointed out that only 24 percent of UK businesses have all five Cyber Essentials controls in place. Heath said the NCSC expects AI to make existing attacks like phishing and payment fraud more effective. He advised companies to focus on security basics and provide clear guidance rather than banning AI tools outright.
New Mexico is moving to protect student data in AI contexts. The Public Education Department proposed rule 6.11.3 NMAC, which would allow vendors to use student data for AI development only if their contracts explicitly permit it. The rule does not ban all AI training with student information. It also clarifies that Amira reading assessment voice recordings do not count as biometric voiceprints. Public comment runs through November 20, 2026, with hearings on November 19 and 20. If adopted, the rule would take effect April 1, 2027.
OpenAI announced it will honor robots.txt signals that let websites opt out of AI training, though the feature is not live yet and no launch date has been set. Meanwhile, Akuna launched a free AI filmmaking academy in Kenya and Nigeria, accepting applications through November 15, 2026. The program covers AI-powered storytelling, visual effects and post-production, with participants working on real-world projects under industry expert feedback.
In other developments, the 2026 World Internet of Things Expo opened in Wuxi, China, drawing over 400 companies from nearly 20 countries across 50,000 square meters. Exhibits featured smart sensors, satellite-based IoT technologies and embodied AI robots. Wuxi's IoT revenue hit 318.4 billion yuan in 2025, and Jiangsu province named the city the core of its three-year IoT strategy. Communication professionals from FIFA World Cup 2026, L'Oreal Paris, USA Today Co. and Bloomberg News discussed AI in creative work at an FIU event, emphasizing that AI should assist rather than replace human storytelling. L'Oreal uses AI for early storyboard exploration, while USA Today Co. built a conversational AI tool for public records requests.
An IDC Analyst Connection report based on a survey of 200 IT and business leaders looked at scaling enterprise AI from pilots to production through private AI clouds. The report identified barriers including data quality and model interpretability, while highlighting benefits such as improved security, compliance and governance.
Key Takeaways
- AI math tutor Goblins improved middle school test scores, linked to about five months of extra learning by Accelerate
- College students saw grades drop by 0.11 points when using a University of Maryland AI assistant, with only 15 percent adoption
- Maple Networks warns employees may accidentally share sensitive business data with AI platforms
- New Mexico proposed rule 6.11.3 NMAC to govern student data use in AI, with comments open through November 20, 2026
- OpenAI will respect robots.txt opt-out signals for AI training, though no launch date is set
- Akuna offers free AI filmmaking training in Kenya and Nigeria through November 15, 2026
- Only 24 percent of UK businesses have all five Cyber Essentials controls, per Simon Heath of The Final Step
- Wuxi, China reported 318.4 billion yuan in IoT revenue in 2025 and hosts the 2026 World IoT Expo with 400+ companies
- L'Oreal Paris and USA Today Co. use AI for storyboard exploration and public records tools respectively, keeping humans in control
- IDC report identifies data quality and model interpretability as barriers to scaling enterprise AI into production
AI tutors help middle schoolers but struggle college students
Two studies by Jing Liu at the University of Maryland show AI tools have very different effects on different students. A math tutor called Goblins helped 6,000 middle schoolers improve their test scores, especially English learners and students with disabilities. But an AI assistant at the University of Maryland hurt 1,100 college students, lowering their grades and reducing their engagement. Only 15 percent of those college students even used the tool. Liu released both studies together to show that AI in education is neither all good nor all bad.
AI tutor boosts math scores but college AI use backfires
Researcher Jing Liu found that an AI math tutor called Goblins improved scores for middle school students who used it at least twice a week. Accelerate, a nonprofit research group, linked the gains to about five months of extra learning. However, the study cannot prove Goblins directly caused the improvement. A separate study found that a University of Maryland AI assistant lowered college students' grades by about 0.11 points. First-generation students were hit hardest by the negative effects. Liu hopes the findings help schools design better AI tools.
Maple Networks warns about AI data security risks
Maple Networks CTO Graham Tetley spoke at International Cyber Expo 2026 about dangers of AI adoption in workplaces. He warned that employees may accidentally share sensitive business data with AI platforms. He said organizations need better visibility into how workers use AI tools. Maple Networks offers 24/7 Security Operations Centre services and CREST-accredited penetration testing to help businesses protect their data. Tetley said preventing unintended data exposure will remain a key priority for security teams.
New Mexico proposes student data privacy rules for AI
New Mexico's Public Education Department proposed rule 6.11.3 NMAC to protect student data and govern contracts with education technology vendors. The rule would allow vendors to use student data for AI development only if their contracts authorize it. It would not ban all AI training with student information. The department also said Amira reading assessment voice recordings do not count as biometric voiceprints under the proposal. Public comment is open through November 20, 2026, with hearings on November 19 and 20. If adopted, the rule would take effect April 1, 2027.
Akuna offers free AI filmmaking training in Kenya and Nigeria
Akuna opened a free AI filmmaking academy for creators in Kenya and Nigeria. Applications are being accepted until November 15, 2026. The program covers AI-powered storytelling, visual effects and post-production techniques. Participants will work on real-world projects and get feedback from industry experts. The goal is to help creators produce high-quality content using AI technology.
OpenAI to honor robots.txt signal for AI training opt out
OpenAI announced it will respect a robots.txt signal that lets websites opt out of AI training. The support is not live yet. No launch date has been given.
World IoT Expo opens in Wuxi with AI and industry focus
The 2026 World Internet of Things Expo opened in Wuxi, China, highlighting the integration of artificial intelligence and IoT. The exhibition covers over 50,000 square meters and has more than 400 companies from nearly 20 countries. Key exhibits include smart sensors, satellite-based IoT technologies, embodied AI robots and next-generation smart devices. Wuxi's IoT revenue reached 318.4 billion yuan in 2025, and a three-year action plan from Jiangsu province identifies the city as the core of its IoT strategy.
Businesses advised to focus on basics for AI security
An interview with Simon Heath of The Final Step highlights that small businesses face AI risks mainly from hype and basic weaknesses. The NCSC expects AI to make existing attacks more effective, yet only 24% of UK businesses have all five Cyber Essentials controls. Key risks include more convincing phishing, payment fraud, AI assistants exposing shared data and unapproved AI tool use by staff. Heath advises getting security basics right and giving clear guidance rather than banning AI.
Communication professionals discuss AI role in creative work
Representatives from FIFA World Cup 2026, L'Oreal Paris, USA Today Co. and Bloomberg News discussed AI use in communications at an FIU event on Oct. 21. Speakers stressed keeping human stories central and using AI as a tool rather than a replacement. L'Oreal uses AI for early storyboard and mood board exploration, USA Today Co. built a conversational AI tool for public records requests, and FIFA's Kristel Valencia-Savage highlighted authenticity as a key human advantage. Students and attendees said the examples helped them see how AI can assist writing while people still guide the process.
IDC report looks at moving enterprise AI to production
An IDC Analyst Connection report examines how private AI clouds can help enterprises scale AI from pilot projects to production. It identifies barriers such as data quality, model interpretability and explainability. The report also notes benefits of private AI clouds, including improved security, compliance and governance. Based on a survey of 200 IT and business leaders, it provides insights into current AI adoption and recommendations for deploying AI at scale.
Sources
- Why AI Tools Work Differently on Middle Schoolers Compared to College Students, According to a Researcher
- Same researcher, opposite results
- Maple Networks Highlights AI Data Security Risks at International Cyber Expo 2026
- New Mexico student privacy proposal would allow some AI training, exclude existing contracts
- Akuna opens free AI filmmaking training to Kenyan and Nigerian creators until 15th November
- OpenAI Announces a robots.txt Signal to Opt Out of AI Training
- World IoT Expo opens in Wuxi, highlighting convergence of artificial intelligence and industry
- How Businesses Can Approach AI Security Without the Technical Jargon: Interview With Simon Heath of The Final Step
- Communication professionals discuss AI’s role in creative work
- IDC Analyst Connection: Moving Enterprise AI from Pilot to Production on a Private AI Cloud
Comments
Please log in to post a comment.